Method of recognizing one or more irradiation

Abstract

Radiologists may protect their subjects against unnecessary exposure to X-rays by use of X-ray opaque material. They may also use X-ray opaque material to permit multiple exposures on a single image. The resulting image contains relatively unexposed regions resulting from the shadow cast by the X-ray opaque material. This invention discloses a method for the automatic determination of the location of the boundary between multiple exposures and the boundary between signal and shadow regions within each exposure. Many hypotheses as to the location of such boundaries are generated and the final decision as to the correct hypothesis is contingent on the results of a rule-based reasoning network.

I claim: 
     
       1. A method of determining a multi-exposure boundary between sub-images in a multi-exposure X-ray image represented by a digital signal representation and of determining a signal/shadow boundary in each sub-image, comprising the steps of i) Extracting low-level line primitives from the X-ray image,   ii) Forming intermediate-level primitives from the low-level primitives,   iii) Building hypotheses as to the location of the multiple-exposure boundary from combinations of intermediate-level primitives,   iv) Associating with each of said hypotheses a cost,   v) Subjecting each of said hypotheses to evaluation tests via a rule-based partitioning reasoning network,   vi) Associating a cost with each of said hypotheses,   vii) Selecting a hypothesis with the least associated cost,   viii) Partitioning the image into sub-images defined by selected hypothesis, and   ix) Submitting each of said sub-images for subsequent detection of a signal/shadow boundary to the following steps (ix) to (xvii):   x) Extracting low-level primitives from each sub-image,   xi) Forming intermediate-level primitives from the low-level primitives extracted from a sub-image,   xii) Building hypotheses as to the location of the signal/shadow boundary from combinations of intermediate-level primitives,   xiii) associating a cost with each of said hypotheses,   xiv) Subjecting each of said hypotheses to evaluation tests via a rule-based collimation reasoning network,   xv) Associating a cost with each of said hypotheses,   xvi) Selecting a hypothesis with the least cost,   xvii) Constructing a binary-valued mask image representing signal and shadow area, said binary-valued mask image being used to extract the signal area in the X-ray image.   
     
     
       2. A method according to claim 1, wherein said low level primitives are lines, said intermediate level primitives are line-groups and wherein said hypotheses for the multiple-exposure boundary are formed from every possible line-cluster and pair of line-clusters, including, a hypothesis that there is no multiple-exposure boundary. 
     
     
       3. A method according to claim 1, wherein the evaluation tests applied to a hypothesis as to the location of the multiple-exposure boundary, comprise any logical combinations of the following tests : --test for boundary closure, which test is passed when line-clusters in a hypothesis under consideration, when having been extrapolated, each terminate on the edge of the image,   --test for boundary location, which test is passed when said line-clusters pass close to the centre of the image,   --test for boundary alignment, which test is passed when said line-clusters are nearly parallel to a side of the image,   --test for boundary completion, which test is passed when the fraction of the boundary perimeter length, for which there exists evidence from the low-level primitives, is greater than a given threshold value.   
     
     
       4. A method according to claim 2 wherein said cost that is assigned to each multiple-exposure hypothesis is lowest if a hypothesis comprises two approximately parallel line-clusters, and wherein said cost increases as the number of visible line-clusters in a hypothesis decreases, and said cost is highest if a hypothesis has no line-clusters. 
     
     
       5. A method according to claim 1 wherein image processing performed on said digital signal representation of said X-ray image is limited to pixel values of said image within a signal/shadow boundary. 
     
     
       6. A method according to claim 1 wherein a digital signal representation of an X-ray image is obtained by scanning an exposed photostimulable phosphor screen with stimulating irradiation, detecting light emitted upon stimulation and converting the detected light into a digital signal representation.